This GitHub project is a port of the `lc` command-line utility from Mark Williams Company's Coherent UNIX. `lc` lists files in categories and columns. This port adds support for symbolic links. It's a handy tool for managing and viewing files.
Read more
This post explores the implementation of Unix pipes in Linux by iteratively optimizing a test program that writes and reads data through a pipe. Starting with a simple program achieving around 3.5GiB/s throughput, the author improves its performance twentyfold through several optimization stages. Key improvements include utilizing `vmsplice` and `splice` system calls to eliminate data copying, leveraging huge pages to reduce paging overhead, and employing busy-looping to minimize synchronization costs. The journey is detailed with code examples and performance analysis using Linux's `perf` tool.
Read more
The Playwright Model Context Protocol (MCP) server provides browser automation capabilities for LLMs using Playwright. It allows LLMs to interact with web pages through structured accessibility snapshots, eliminating the need for screenshots or visually-tuned models. It's fast, lightweight, and LLM-friendly, using Playwright's accessibility tree rather than pixel-based input. Features include web navigation, form filling, data extraction, and automated testing. Supports headless and headed modes. Installation is straightforward via VS Code CLI.
Read more
Evelyn, a Los Angeles woman, lost $300,000 to a romance scam orchestrated through the Hinge dating app. The scammer, posing as "Bruce," lured her into a cryptocurrency investment scheme, ultimately stealing her life savings. This case highlights the growing use of AI in scams: AI writing tools make it easier to create convincing narratives, while deepfakes enhance credibility, making scams harder to detect. Evelyn's story serves as a cautionary tale, emphasizing the importance of caution in online dating and the dangers of high-yield investment promises.
Read more
New York's Affordable Broadband Act, requiring ISPs to offer low-cost plans to low-income residents, is now in effect after a lengthy legal battle. Following a Supreme Court decision against industry challenges, the law mandates $15 or $20 monthly plans with varying speeds. The law aims to bridge the digital divide, particularly after a federal program expired, leaving millions without subsidized internet. Small ISPs may seek exemptions, while larger providers face penalties for non-compliance.
Read more
This GitHub repository, awesome-donations, is a curated list of donation options for numerous Free/Libre and Open Source Software (FLOSS) projects. It provides easy access to support projects ranging from large foundations like the Linux Foundation and Mozilla Foundation to individual projects such as LibreOffice, GIMP, and QEMU. Whether you prefer PayPal, credit cards, or other methods, you can easily contribute to the open-source community and help ensure the continued development of essential free software.
Read more
After a year and a half of development, the author proudly presents Game Bub, an open-source FPGA-based retro gaming handheld supporting Game Boy, Game Boy Color, and Game Boy Advance games. This detailed write-up chronicles the journey, from initial concept to final assembly, including hardware selection (a Xilinx XC7A100T FPGA at its core), PCB design, firmware development in Rust and Slint, and 3D-printed enclosure creation. Game Bub plays both physical cartridges and ROMs from a microSD card, and even boasts features like HDMI output, rumble, and a real-time clock. The project showcases a remarkable feat of engineering and a dedication to open-source principles.
Read more
FlashMLA is a highly efficient MLA decoding kernel optimized for Hopper GPUs, designed for variable-length sequence serving. Achieving up to 3000 GB/s in memory-bound configurations and 580 TFLOPS in computation-bound configurations on H800 SXM5 using CUDA 12.6, FlashMLA utilizes BF16 precision and a paged kvcache with a 64 block size. Inspired by FlashAttention 2&3 and the cutlass projects, FlashMLA offers significant performance improvements for large-scale sequence processing.
Read more
This article critically examines Apple CEO Tim Cook's relationship with the Trump administration. The author argues that Cook, in pursuit of tax and tariff advantages for Apple, repeatedly ingratiated himself with Trump, including donating to his inauguration and meeting with other CEOs. This, the author contends, directly contradicts Cook's public image as a supporter of the LGBTQ+ community, especially given the Trump administration's anti-LGBTQ+ policies. The author calls for critical reflection on Cook's actions and urges Cook to leverage his influence to protect the rights of his employees and customers, rather than solely pursuing corporate gain.
Read more
The Recursive Book of Recursion challenges the intimidating reputation of recursive algorithms. Using Python and JavaScript examples, it progressively explains recursion's fundamentals, common algorithms (factorials, Fibonacci sequences, tree traversal, maze solving, etc.), and optimization techniques (tail call optimization, memoization). It covers advanced topics like divide-and-conquer, permutations, combinations, and dynamic programming, culminating in fractal art generation. Project-based and highly accessible, this book is perfect for beginners and experienced programmers alike.
Read more
NetBox Labs recently released a public preview of its NetBox Discovery agent. This fully open-source tool quickly and easily discovers networks and devices, ingesting information into NetBox to accelerate building a network source of truth centered around NetBox. Its agent-based architecture is ideal for complex network environments and works with NetBox Assurance to detect and remediate network drift. Currently supporting two discovery modes: network and device discovery, it integrates with the Diode data ingestion engine.
Read more
Researchers introduce SGD-SaI, a novel optimizer improving stochastic gradient descent. SGD-SaI addresses training imbalances by scaling learning rates at initialization for different parameter groups based on their gradient signal-to-noise ratios. Significantly more memory-efficient than AdamW, SGD-SaI matches or surpasses AdamW's performance across various Transformer-based tasks, including ImageNet classification and LLM pretraining. Its robustness and practicality are demonstrated across diverse applications, making it a compelling alternative.
Read more
This article highlights a common C++ pitfall: accidentally copying data instead of referencing it due to a missing ampersand (&). The author demonstrates how Rust's default move semantics and borrow checker prevent this subtle but performance-critical error. Using examples like `vec::retain`, the article shows how Rust's compiler proactively catches such mistakes at compile time, enhancing code reliability. While C++ offers mechanisms to avoid this, Rust's defaults are simpler and more effective, reducing cognitive load for developers. The comparison also touches upon idiomatic versus unidiomatic Rust code, showing that even non-conventional approaches are less prone to these errors in Rust.
Read more
This blog post, part four in a series on optimizing Ruby's JSON performance, details the author's journey in improving Ruby's JSON serialization speed. Through meticulous micro-benchmarking and profiling, the author explores stack allocation and inlining techniques. By shifting buffer allocation from the heap to the stack and strategically using inlining, significant performance gains are achieved. However, the article highlights the importance of balancing micro-benchmark improvements with real-world application performance, showcasing a case where optimization negatively impacted larger datasets.
Read more
Apple announced the iPhone Air, its thinnest iPhone yet, boasting a groundbreaking titanium design and pro-level performance. Powered by the A19 Pro, N1, and C1X chips, it features a stunning 6.5-inch Super Retina XDR display, a versatile 48MP Fusion camera system, and all-day battery life. Pre-orders start September 12th, with availability beginning September 19th, starting at $999.
Read more
Oracle co-founder Larry Ellison envisions a future where AI powers a massive surveillance system, constantly monitoring citizens to ensure "good behavior." This system would analyze data from various sources, including police body cameras, security cameras, and even doorbell cameras. Ellison also predicts AI drones replacing police cars in high-speed chases. While highlighting potential improvements in policing, his comments raise significant concerns about privacy and the potential for societal control.
Read more
OpenAI has found evidence suggesting that Chinese AI company DeepSeek used OpenAI's model data to train its own low-cost AI models, potentially violating its terms of service. DeepSeek allegedly employed a 'distillation' technique to extract data from OpenAI's models, enabling it to train its own models at a fraction of the cost—far less than the $100 million OpenAI spent on GPT-4. OpenAI and Microsoft are investigating the matter, sparking a debate about AI intellectual property and data security, and highlighting the intensifying competition among tech giants.
Read more
Boeing reported a $523 million loss on its CST-100 Starliner program in 2024, bringing the total losses to over $2 billion. The losses are attributed to schedule delays, increased testing and certification costs, and higher post-certification mission costs. While NASA reported progress on resolving issues from a previous test flight, key problems, including thruster malfunctions, remain unresolved. The timeline for Starliner's next flight, whether crewed or uncrewed, remains uncertain.
Read more
The National Security Archive has released declassified documents detailing the 'Nth Country Experiment,' a secret mid-1960s project at Lawrence Livermore National Laboratory. A small team of physicists, using only publicly available information, designed a functional nuclear weapon in just three years. This 'do-it-yourself' project demonstrated the feasibility of nuclear weapon development with limited resources, highlighting the dangers of nuclear proliferation. The released documents, while heavily redacted, reveal insights into the experiment's methodology and conclusions, sparking renewed discussion about the protection of nuclear weapons design information and the threat of nuclear terrorism.
Read more
The International Obfuscated C Code Contest (IOCCC) is holding its 28th competition, marking its 40th anniversary. This event celebrates creativity and quirky coding skills within the C programming language. To use the Mastodon web application, JavaScript is required; alternatively, consider using a native Mastodon app.
Read more
New research reveals a surprising finding: people with less knowledge about AI are more open to integrating it into their daily lives. This contradicts common assumptions. The study found higher AI acceptance rates in nations with lower average AI literacy. The reason? AI's ability to perform tasks previously thought exclusive to humans creates a sense of wonder and awe. Those familiar with AI's technical workings see it as a tool, not magic. Promoting AI requires balancing public understanding with maintaining enthusiasm to fully harness its potential.
Read more
Finley Technologies, a Y Combinator and Bain Capital Ventures-backed fintech startup, is seeking a Growth Associate. This role focuses on the credit fund segment and requires 2-3 years of finance experience, strategic thinking, entrepreneurial spirit, and strong communication skills. The successful candidate will help shape go-to-market strategy, product roadmap, and more, collaborating with a team to drive company growth.
Read more
ITXPlus is a Mini-ITX sized Macintosh Plus logic board clone built entirely from modern components. It uses a Pico-based VGA converter, a standard 24-pin ATX power supply, an onboard 50-pin internal SCSI header, and 4MB of soldered RAM. Leveraging open-source solutions like DosFox's sound chip replacement and an ATTiny-based RTC, ITXPlus aims for a fully functional, build-it-yourself experience. While floppy drive support isn't included by default, an expansion header allows for the addition of a real IWM. The board is almost entirely surface mount, using a traditional 68000 for easier assembly. The complete design will be open-source on GitHub.
Read more
System76 released COSMIC Alpha 6, a significant update to its Rust-based desktop environment. New features include Desktop Zoom for easy scaling, improved workspace management with intuitive window switching and dragging, and enhancements to the file manager, media player, and text editor. Numerous bug fixes boost performance and stability. Alpha 6 also optimizes fonts, memory management, and CPU usage for a smoother experience.
Read more
This article explores the complex relationship between humans and rats. Historically demonized for spreading diseases like the plague, recent research suggests that rats may not be solely to blame for plague transmission; human hygiene plays a crucial role. The article further reveals that urban rats aren't as dirty or aggressive as commonly perceived, nor are they superspreaders of disease. Instead of a brutal war on rats, the author calls for communication and coexistence, suggesting improvements to infrastructure, sanitation, and other measures to foster a more harmonious relationship with rats.
Read more
BioNTech presented early clinical trial data for its novel bispecific antibody, BNT-327, at the San Antonio Breast Cancer Symposium. Targeting PD-1/PD-L1 and VEGF, the antibody showed positive results in patients with triple-negative breast cancer. Building on the discovery of checkpoint inhibitors, this research represents a potential breakthrough in next-generation immunotherapy, offering hope for new treatments in triple-negative breast cancer and potentially other cancers.
Read more
Maciej "Groobo" Maselewski reigned supreme in Diablo speedrunning for years, his 3-minute, 12-second Sorceror run seemingly unbeatable. However, a team of speedrunners, attempting to replicate his seemingly lucky dungeon runs using external software, uncovered inconsistencies. An automated search through billions of legitimate Diablo dungeons proved Groobo's run impossible within the game's legitimate parameters. This revelation sparked controversy within the speedrunning community, exposing years of unearned praise and accolades based on fraudulent gameplay.
Read more
Researchers discovered two critical vulnerabilities, dubbed FLOP and SLAP, in Apple's M-series and A-series chips. FLOP exploits the chip's load value predictor (LVP) to steal sensitive data from Chrome and Safari browsers, including information from Gmail, iCloud, and Google Maps. SLAP, targeting primarily Safari, leverages the load address predictor (LAP) for similar data theft. Affected devices include iPhones, iPads, and Macs released since September 2021. While Apple claims to be assessing the risk, researchers have published mitigations and recommend users update their systems.
Read more
Traditional time-series forecasting methods like ARIMA and Prophet are being challenged by a new generation of "foundation models." These models aim to bring the power of large language models (LLMs) to time-series data, enabling a single model to forecast across diverse datasets and domains. This article benchmarks several foundation models—Amazon Chronos, Google TimesFM, IBM Tiny Time-Mixers, and Datadog Toto—against classical baselines. Testing on real-world Kubernetes pod metrics reveals that foundation models excel at multivariate forecasting, with Datadog Toto performing particularly well. However, challenges remain in handling outliers and novel patterns, and classical models retain competitiveness for steady-state workloads. Ultimately, the authors conclude that foundation models offer significant advantages for fast-changing, multivariate data streams, providing more flexible and scalable solutions for modern observability and platform engineering teams.
Read more
Stripe recently experienced an unexpected spike in DNS errors. This post details how they used tools like Unbound, tcpdump, and iptables to track down the root cause. The investigation revealed that a Hadoop job analyzing network logs was performing numerous reverse DNS lookups (PTR records), leading to traffic amplification due to retries exceeding the AWS VPC resolver's limits. Stripe resolved the issue by adjusting Unbound forwarding configurations to distribute the load across individual Hadoop hosts. The case highlights the importance of robust monitoring, multi-faceted troubleshooting, and strategies for handling traffic surges in high-availability systems.
Read more